Innovation Profs - 9/17/2024

New half-day workshops

Join the Innovation Profs for two new half-day, virtual workshops on generative AI tools:

Oct. 2 Intro to Generative AI: This virtual workshop will help those new to generative AI get quickly up to speed on the tools available, their capabilities and their limitations.

Oct. 16 Using Gen AI at Your Work: This completely new virtual workshop will take a hands-on approach to helping you get implement generative AI at your work. You’ll learn our process to determine quick wins, imagine new ideas and prepare for the future of AI at your company.

Generative AI News

OpenAI releases o1, its first model with ‘reasoning’ abilities

Strawberry is finally here! Rumors of OpenAI’s new large language model have been floating around since earlier this summer, and now, the rumors are reality: On Friday, OpenAI released GPT-4o1 (in a preview form) as well as the smaller and cheaper GPT-4o1 mini, both of which are models that have been specifically developed to handle more complex reasoning tasks. According to OpenAI’s announcement of the new models, “We trained these models to spend more time thinking through problems before they respond, much like a person would. Through training, they learn to refine their thinking process, try different strategies, and recognize their mistakes.” Exactly how these models work and were trained is unclear; OpenAI’s head of research Jerry Tworek told the Verge that the model GPT-4o1 “has been trained using a completely new optimization algorithm and a new training dataset specifically tailored for it.” This apparent improvement in performance comes at a cost, however, as GPT-4o1-preview is 3-4 times more expensive (measured in terms of cost per one million tokens) compared to GPT-4o.

Microsoft Copilot's Wave 2 is here. Everything you need to know

On Monday, Microsoft unveiled new features of their generative AI platform Copilot in a live-streamed event called “Microsoft 365 Copilot: Wave 2.” Some of these new features include:

• Copilot Pages, which Microsoft’s describes as “a dynamic, persistent canvas designed for multiplayer AI collaboration”

• Copilot in the Microsoft 365 apps, including Teams, Excel, Powerpoint, Word, Outlook, and OneDrive

• Copilot with Python in Excel, which allows users to perform more advanced analyses in Excel using Python, mediated through a natural language conversation, i.e., in a code-free fashion

• Copilot agents, which can be used to automate a range of processes (not to be confused with copilots, which work in tandem with human users)

This is certainly the most significant development to date in assimilating generative AI into an existing suite of productivity tools.

Google's DataGemma is the first large-scale Gen AI with RAG - why it matters

Late last week, Google announced DataGemma, a family of open-source models that use a RAG (retrieval augmented generation) framework with Google’s Data Commons project to ground LLM responses in publicly available data, with the aim of minimizing hallucinations. “While RAG is becoming a popular approach for enabling enterprises to ground LLMs in their proprietary corporate data, using Data Commons represents the first implementation to date of RAG at the scale of cloud-based Gen AI.” In addition, DataGemma implements a new technique called RIG (retrieval interleaved generation), which involves the incorporation of data retrieved from Data Commons as the model is producing its response.

Quick Hits

Tool of the week: Google Vids

Google Vids is an experimental tool from Google that uses AI to create videos from your Google Docs and Slides. Google is currently testing this new application with a select group of trusted testers in Google Workspace Labs.

Learn more about Google Vids.

AI-generated image of the week

For a presentation we gave earlier this week, we needed an image to transition from text-based generative AI to image-based. This wasn’t the one we chose, but we still liked the way it turned out.

prompt: a photo that is made out of words

Get starting with Generative AI

New to generative AI? Here are some places to start…

• NightCafe image prompting tips

• Learn ChatGPT prompting

• Best AI image generators

• AI terms everyone should know

• Prompt Engineering: What to Know and Why It's Important 

• 7 amazing Midjourney styles to bring your AI images to life

What we found

Adobe gave a sneak peek of its upcoming generative AI video tools, featuring a new capability that can create video clips from still images. This preview expands on the Firefly video model, first showcased in April, which will soon power AI-driven video and audio editing tools within Adobe's Creative Cloud applications.